[Mailman-Developers] suggested improvement for Mailman's bounce processing

James Ralston qralston+ml.mailman-developers at andrew.cmu.edu
Mon Aug 7 21:08:56 CEST 2006


On 2006-07-28 at 21:31-05 Brad Knowles <brad at stop.mail-abuse.org> wrote:

> Unfortunately, there are a whole host of seriously broken MTAs out
> there, and seriously broken configurations of otherwise good MTAs,
> and many sites return totally bogus status codes.  In many cases,
> site admins will blindly copy stuff from somewhere else that was
> horribly broken to begin with and won't understand what's wrong with
> it before they do the cut-n-paste operation.

Perhaps, but we cannot solve this problem, and there's a fine line
between working around stupidity and coddling it.

> That said, I would not be opposed to seeing more data on this
> subject, and possibly giving site admins or list admins an option
> they can enable that would allow Mailman to pay attention to the
> status codes.  Once that's out there, we could let various people
> try it out and see how it works in the field, and I would be a very
> happy guy if I were to be proven wrong in this case.

What further data do you wish to see?  I think I've documented the
problem well enough.  There's no way we know many horribly broken
sites are out there.

On 2006-07-31 at 10:56+01 Ian Eiloart <iane at sussex.ac.uk> wrote:

> I don't see how that could create a problem.  The worst thing that
> could happen is that someone remains subscribed to a list when they
> should not.  Alternately, they could be unsubscribed because their
> MTA is returning the wrong error codes - but then that would give
> their postmaster a good reason to fix the error codes.  In this
> case, they'd be unsubscribed as things stand anyway.

Right: the only risk is that bounces coming from a subscriber at a
broken site might be ignored, because they look like they're being
generated based on the content of certain messages.

IMHO, this risk is negligible.  If the operators of the broken site in
question get annoyed that Mailman keeps trying to send messages to a
non-existent address, they should fix their broken site.

As a compromise, I suggest adding this feature as a bounce processing
tunable; for example, "content bounce handling":

    Setting:

        How should Mailman handle bounces that appear to be related to
        content?

    Description:

        Sometimes a message to a subscriber bounces due to the content
        of the message, not because the subscriber's address is
        invalid.  This option controls how Mailman handles bounces
        that appear to be related to the content of messages.

        Picking "count the bounce" will cause Mailman to count any
        bounce against the bounce threshold, regardless of the reason
        why the message bounced.  The advantage of this option is that
        it is least likely to "miss" bounces.  The disadvantage of
        this option is that it penalizes subscribers at sites that
        correctly indicate why a message bounced.

        Picking "forward the bounce to the list owner" will cause
        Mailman to forward bounces that seem to be related to the
        content of specific messages to the list owner.  The advantage
        of this option is that the owner will be able to review the
        bounce and take appropriate action; the disadvantage is that
        the list owner might be overwhelmed by bounces.

        Picking "ignore the bounce" will cause Mailman to ignore
        bounces that appear to be related to the content of specific
        messages.  The advantage of this option is that subscribers at
        sites that correctly indicate why a message bounced won't be
        penalized.  The disadvantage is that if a misconfigured site
        erroneously indicates that *all* messages are due to content,
        then Mailman will never detect bouncing subscribers at that
        site.

    Choices:

        [X] Count the bounce against the threshold.
        [ ] Forward the bounce to the list owner.
        [ ] Ignore the bounce.

Comments?

James



More information about the Mailman-Developers mailing list